Table 1.
Publications describing the application of machine learning approaches to neoepitope prediction.
| Publication | Indication | Sample type and number | number of HLAs used | Estimated ratio of predicted neoepitopes from mutations | Estimated ratio of experimentally confirmed neoantigens | Number of features | Algorithms |
|---|---|---|---|---|---|---|---|
| (Segal et al., 2008) | BRCA/CRC | 11 patients | 1 | 0.17 | N/A | 1 | NetMHC, SYFPEITHI, BIMAS, RANKPEP |
| (Castle et al., 2012) | MEL | 1 murine cell line | N/S | 0.05 | 0.32T | 2 | NetMHC |
| (Khalili et al., 2012) | various | 312 genes (COSMIC) | 57 | 1.40 | N/A | 2 | NetMHC 3.2 |
| (Robbins et al., 2013) | MEL | 3 patients | 2 | 0.18 | 0.03 T | 3 | NetMHCpan 2.4 |
| (van Rooij et al., 2013) | MEL | 1 patient | 4 | 0.42 | <0.01 T | 3 | NetChop, NetMHC 3.2 |
| (Boegel et al., 2014) | various | 167 cancer cell lines | 6 | 0.44 | N/A | 1 | IEDB 2.9 |
| (Duan et al., 2014) | SARC | 2 murine tumors | 3 | 0.75 | 0.56 T | 2 | NetMHC 3.0 |
| (Snyder et al., 2014) | MEL | 64 patients | 6 | 0.42 | <0.01 T | 3 | NetMHC 3.4, RANKPEP, IEDB immunogenicity, CTLPred |
| (Yadav et al., 2014) | CRC/PRAD | 2 murine cell lines | 2 | 0.03 | 0.02 T | 3 | NetMHC 3.4 |
| (Angelova et al., 2015) | CRC | 552 TCGA patients | 6 | 0.41 | N/A | 2 | NetMHCpan |
| (Carreno et al., 2015) | MEL | 7 samples/3 patients | 1 | 0.04 | 0.43 B | 3 | NetMHC 3.4 |
| (Cohen et al., 2015) | MEL | 8 patients | 2 | 0.02 | 0.02 T | 2 | IEDB |
| (Rizvi et al., 2015) | NSCLC | 34 patients | 6 | 0.62 | <0.01 T | 2 | NetMHC 3.4 |
| (Rooney et al., 2015) | various | 4250 TCGA patients | 6 | 0.14 | N/A | 2 | NetMHCpan 2.4 |
| (Tran et al., 2015) | GIC | 10 patients | 12 | 0.03 | 0.21 T | 2 | NetMHCpan 2.8, NetMHCIIpan 3.0 |
| (Van Allen et al., 2015) | MEL | 110 patients | 6 | 1.56 | N/A | 2 | NetMHCpan 2.4 |
| (van Gool et al., 2015) | UCEC | 245 TCGA patients | 1 | 0.06 | N/A | 3 | NetMHCpan 2.8 |
| (Bassani-Sternberg and Gfeller, 2016a) | MEL | 1 patient | 6 | 1.43 | <0.01 B | 1 | NetMHCpan 2.8 |
| (Goh et al., 2016) | MCC | 49 patients | 4 | 0.09 | N/A | 1 | NetMHC 3.4 |
| (Gros et al., 2016) | MEL | 3 patients | 6 | 0.03 | 0.55 T | 2 | IEDB |
| (Hugo et al., 2016) | MEL | 38 patients | 12 | 0.06 | N/A | 3 | NetMHCpan 2.8, NetMHCIIpan 3.0 |
| (Kalaora et al., 2016) | MEL | 1 patient | 6 | 5.30 | <0.01 B | 1 | NetMHCpan 2.8 |
| (Karasaki et al., 2016) | NSCLC | 15 patients | 6 | 0.62 | N/A | 1 | NetMHCpan 2.8 |
| (Löffler et al., 2016) | CHOL | 1 patient | 6 | 3.68 | 0 B | 2 | NetMHC 3.4, NetMHCpan 2.8, SYFPEITHI |
| (Strønen et al., 2016) | MEL | 3 patients | 1 | 0.05 | 0.19 T | 4 | NetChop, NetMHC 3.2, NetMHCpan 2.0 |
| (Anagnostou et al., 2017) | NSCLC | 10 patients | 6 | 0.76 | <0.01 T | 4 | SYFPEITHI, NetMHCpan, NetCTLpan |
| (Chang et al., 2017) | PED | 540 patients | 6 | 0.42 | N/A | 2 | NetMHCcons 1.1 |
| (Karasaki et al., 2017) | NSCLC | 4 patients | 6 | 0.20 | N/A | 2 | NetMHCpan 2.8 |
| (Kato et al., 2017) | BRCA | 5 patients | 6 | 0.47 | N/A | 2 | NetMHC 3.4, NetMHCpan 2.8 |
| (Miller et al., 2017) | MM | 664 patients | 6 | 0.16 | N/A | 3 | NetMHC 4.0 |
| (Ott et al., 2017) | MEL | 6 patients | 6 | 0.01 | 0.60 T | 3 | NetMHCpan 2.4 |
| (Sahin et al., 2017) | MEL | 13 patients | 10 | 0.02 | 0.60 T | 2 | IEDB 2.5 (MHC class I & II) |
| (Zhang et al., 2017) | BRCA | 3 patients | 6 | 0.01 | 0.16 T | 3 | NetMHC 3.2 |
| (Kalaora et al., 2018) | MEL | 15 patients/cell lines | 6 | 9.57 | 0.15 T | 2 | NetMHCpan 3.0 |
| (Kinkead et al., 2018) | PAAD | 1 murine cell line | 2 | 0.27 | 0.16 T | 2 | NetMHC 3.2/3.4, NetMHCpan 2.8 |
| (Martin et al., 2018) | OV | 1 patient | 6 | 1.57 | 0,09 T | 2 | NetMHCpan 2.4 |
| (O’Donnell et al., 2018a) | OV | 92 patients | 6 | 0.02 | N/A | 2 | NetMHCpan 2.8 |
| (Sonntag et al., 2018) | PDAC | 1 patient | 10 | 2.00 | 0.75 T | 3 | NetMHC, NetMHCIIpan 3.1, SYFPEITHI |
| (Thorsson et al., 2018) | various | 8546 TCGA patients | 6 | 0.74 | N/A | 2 | NetMHCpan 3.0, pVAC-Seq 4.0.8 |
| (Vrecko et al., 2018) | HCC | 1 patient | 3 | 0.05 | 0.15 T | 2 | SYFPEITHI, IEDB (MHC class II) |
| (Wu et al., 2018) | various | 7748 TCGA samples | 100 | 1.18 | N/A | 1 | NetMHCpan 4.0 |
| (Bulik-Sullivan et al., 2019) | NSCLC | 7 patients | 6 | 0.10 | 0.08 T | >4 | EDGE |
| (Hilf et al., 2019) | GBM | 10 patients | 1 | 0.03 | 0.85 T | 3 | IEDB 2.5 |
| (Keskin et al., 2019) | GBM | 8 patients | 6 | 0.20 | 0.07 T | 3 | NetMHCpan 2.4 |
| (Koster and Plasterk, 2019) | various | 10186 TCGA patients | 1 | 0.02 | N/A | 2 | NetMHC 4.0 |
| (Liu et al., 2019) | OV | 20 patients | 12 | 0.15 | 0.24 T | 3 | NetMHCpan 3.0, NetMHCIIpan 3.1 |
| (Löffler et al., 2019) | HCC | 16 patients | 6 | 1.79 | 0 B | 2 | NetMHC 4.0, NetMHCpan 3.0, SYFPEITHI |
| (Rosenthal et al., 2019) | NSCLC | 164 samples/64 patients | 6 | 0.86 | N/A | 2 | NetMHC 4.0, NetMHCpan 2.8 |
| (Schischlik et al., 2019) | PNMN | 113 patients | 6 | 2.53 | 0.66 B | 2 | NetMHCpan |
N/S means not specified. Cancer type abbreviations: adenocarcinoma (AC), breast cancer (BRCA), cholangiocarcinoma (CHOL), colorectal cancer (CRC), glioblastoma (GBM), gastrointestinal cancer (GIC), hepatocellular carcinoma (HCC), merkel cell carcinoma (MCC), melanoma (MEL), multiple myeloma (MM), non-small cell lung cancer (NSCLC), ovarian cancer (OV), pancreatic ductal adenocarcinoma (PDAC), pediatric cancers (PED), Ph-negative myeloproliferative neoplasms (PNMN), prostate adenocarcinoma (PRAD), sarcoma (SARC) and uterine corpus endometrial cancer (UCEC). T indicates experimentally confirmed T cell responses (e.g., IFNγ ELISPOT), B indicates experimentally confirmed major histocompatibility complex (MHC) binding (e.g., mass spectrometric [MS] of eluted peptides), and N/A indicates that no experimental validation was done. Features are mutated peptide binding prediction, wild-type peptide binding prediction, gene expression, sequence-based features like sequence similarity scores, and immunogenicity predictions. If available, version information of algorithms is included.